[DSC Europe 22] Next-Wave of Value – Operating Model for Scaling Data Science & AI - Michael Soucek
Download as PPTX, PDF0 likes76 views
The document discusses the key factors for successfully scaling AI within organizations, emphasizing the importance of a defined strategy, multi-disciplinary teams, and a strong data foundation. Despite 63% of firms advancing their AI maturity, many are still not realizing AI's full potential due to poor organizational setups and a lack of responsibility for scaling. Successful companies have created 'AI factories' that prioritize flexibility, cost efficiency, and integration with enterprise strategies from the outset.
[DSC Europe 22] Next-Wave of Value – Operating Model for Scaling Data Science & AI - Michael Soucek
- 1. Next-Wave of Value – Operating Model for Scaling Data Science & AI Dr. Michael Soucek Head of Data Science Accenture AI ASG 17.11.2022
- 2. “I feel like having more (AI) pilots than any airline – still I don’t get anywhere” CEO German chemicals company
- 3. (1) State of the Nation (2) What we know succesful AI scaler have in common? (3) What do they specifically look at in their operating model?
- 4. In fewer than 70 years, artificial intelligence (AI) has evolved from a scientific concept to a societal constant of firms are still testing the AI waters 63% of firms have advanced their AI maturity enough to achieve superior performance and growth 12% When it comes to making the most of AI’s full potential, most organizations are barely scratching the surface… Quelle: Der KI-Reifegrad | Accenture
- 5. AI maturity* measures the degree to which organizations have mastered AI-related capabilities… You know what, and don‘t know how to scale. You are not sure what to do and how. You could do it, but dont`know what. *Accenture Research analysis based on a sample of 1,200 companies
- 6. … and it strongly varies by industry *Accenture Research analysis based on a sample of 1,200 companies
- 7. What we know from successful AI scalers: AI is everyone’s business. have a clearly- defined strategy and operating model for scaling AI 71% leverage multi-disciplinary teams 92% agree a core data foundation is crucial to scaling AI 72% more likely to understand and implement Responsible AI 1.7x more likely to have employees that understand how AI applies to their role 2x
- 8. If most organizations are racing to embrace AI, why are some seeing more value than others? Value + Strategy Their top leaders champion AI as a strategic priority for the entire organization People & Capabilities They invest heavily in talent to get more from their AI investments Governance They design AI responsibly, from the start. They industrialize AI tools and teams to create a strong AI core Optimization & scale Investment Priority They prioritize long- and short-term AI investment
- 9. Let‘s be a little more concrete!
- 10. Organizational development of both achievers and experimenters is similar Beginning of Data & Analytics Journey Thin Hub, Thick Spokes Thick Hub, Thin Spokes Balanced Corporate Centralized Model End of Data & Analytics Journey … I am sure you have heard about this plan!
- 11. The devil is in detail CoE Business „We spend huge amout of time scanning and undestanding analyst reports and often miss important topics. Anythig AI can help with? Nr. 1 „We can train BERT based models to undestand entity linking in written text? ? Nr. 2 „Sure, I did something similar in R&D of a Pharma Comapny“ Difficult experience of business stakeholders with early hubs… CoE The reason is often missing or wrongly placed “Jedis”! „No time fot this, I dont`s see tha value“ „I don‘t understand what they are saying. Let me hire my own team“ … have negative impact on the acceptance of AI in the organization
- 12. In case the first step works, it often fails after the pilot Lacking responsibility and capability to scale… … is the real AI innovation blocker for most of the organizations. CoE Business „That really cool idea and the modell seems to work. Lets roll out to other countries for different languages! Can we do this? What does it cost? ? „Let`s talk to the IT...“ CoE IT & Platform Team “Even if we can set it up as an IT Service, who is going to deliver and maintain the solution?” „ This deos not scale“ „We are more an enabler“ The reason is missing platform, processes, right people and data governance.
- 13. Missing capability to scale... … a bit of the “Czech Dream”.
- 14. Lab / Factory Successful companies have design AI factories Analytics Factory Analytics Lab Information Management 1) Key role of Analytics Translators who are close to business (often in the business domains) 2) Data Science Teams often dedicated to business domains 3) Hubs are becoming flexible, cost-efficient factories with multi-skill POD Teams 4) Hubs are build close to the plattform in alignment with enterprise deployment strategy 5) Data Governance has high priority Analytics Platform Delivery Business oriented data science teams, Translators, data citizens and data owners They were building AI Factories from day 1… … while many were focusing only on the Labs. What are the common themes we have observed?
- 15. 3 Takeaways (1) Understand you Pain Points: Massive potential is still left on the way mainly due to wrong organizational set up (2) You need the right talent: Success is a play of strategy, business and plattform – right people for the right job (3) Build Your Factory: Hubs should have a real AI Factories which are agile, flexible, cost efficient and close to data and platform
- 16. Danke!
Editor's Notes
- #15: Proof of Concept Industrialization Embed insights into decision making process Harden and automate the data supply chain Roll out of the solution across the priority markets Deploy on platform allowing for visualization and data exploration by Data Consumers